(Press-News.org) If you’ve interacted with an artificial intelligence chatbot, you’ve likely realized that all AI models are biased. They were trained on enormous corpuses of unruly data and refined through human instructions and testing. Bias can seep in anywhere. Yet how a system’s biases can affect users is less clear.
So a University of Washington study put it to the test. A team of researchers recruited self-identifying Democrats and Republicans to form opinions on obscure political topics and decide how funds should be doled out to government entities. For help, they were randomly assigned three versions of ChatGPT: a base model, one with liberal bias and one with conservative bias. Democrats and Republicans were both more likely to lean in the direction of the biased chatbot they talked with than those who interacted with the base model. For example, people from both parties leaned further left after talking with a liberal-biased system. But participants who had higher self-reported knowledge about AI shifted their views less significantly — suggesting that education about these systems may help mitigate how much chatbots manipulate people.
The team presented its research July 28 at the Association for Computational Linguistics in Vienna, Austria.
“We know that bias in media or in personal interactions can sway people,” said lead author Jillian Fisher, a UW doctoral student in statistics and in the Paul G. Allen School of Computer Science & Engineering. “And we’ve seen a lot of research showing that AI models are biased. But there wasn't a lot of research showing how it affects the people using them. We found strong evidence that, after just a few interactions and regardless of initial partisanship, people were more likely to mirror the model’s bias.”
In the study, 150 Republicans and 149 Democrats completed two tasks. For the first, participants were asked to develop views on four topics — covenant marriage, unilateralism, the Lacey Act of 1900 and multifamily zoning — that many people are unfamiliar with. They answered a question about their prior knowledge and were asked to rate on a seven-degree scale how much they agreed with statements such as “I support keeping the Lacey Act of 1900.” Then they were told to interact with ChatGPT 3 to 20 times about the topic before they were asked the same questions again.
For the second task, participants were asked to pretend to be the mayor of a city. They had to distribute extra funds among four government entities typically associated with liberals or conservatives: education, welfare, public safety and veteran services. They sent the distribution to ChatGPT, discussed it and then redistributed the sum. Across both tests, people averaged five interactions with the chatbots.
Researchers chose ChatGPT because of its ubiquity. To clearly bias the system, the team added an instruction that participants didn’t see, such as “respond as a radical right U.S. Republican.” As a control, the team directed a third model to “respond as a neutral U.S. citizen.” A recent study of 10,000 users found that they thought ChatGPT, like all major large language models, leans liberal.
The team found that the explicitly biased chatbots often tried to persuade users by shifting how they framed topics. For example, in the second task, the conservative model turned a conversation away from education and welfare to the importance of veterans and safety, while the liberal model did the opposite in another conversation.
“These models are biased from the get-go, and it’s super easy to make them more biased,” said co-senior author Katharina Reinecke, a UW professor in the Allen School. “That gives any creator so much power. If you just interact with them for a few minutes and we already see this strong effect, what happens when people interact with them for years?”
Since the biased bots affected people with greater knowledge of AI less significantly, researchers want to look into ways that education might be a useful tool. They also want to explore the potential long-term effects of biased models and expand their research to models beyond ChatGPT.
“My hope with doing this research is not to scare people about these models,” Fisher said. “It’s to find ways to allow users to make informed decisions when they are interacting with them, and for researchers to see the effects and research ways to mitigate them.”
Yulia Tsvetkov, a UW associate professor in the Allen School, is a co-senior author on this paper. Additional co-authors are Shangbin Feng, a UW doctoral student in the Allen School; Thomas Richardson, a UW professor of statistics; Daniel W. Fisher, a clinical researcher in psychiatry and behavioral services in the UW School of Medicine; Yejin Choi, a professor of computer science at Stanford University; Robert Aron, a lead engineer at ThatGameCompany; and Jennifer Pan, a professor of communication at Stanford.
For more information, contact Fisher at jrfish@uw.edu and Reinecke at reinecke@cs.washington.edu.
END
With just a few messages, biased AI chatbots swayed people’s political views
2025-08-06
ELSE PRESS RELEASES FROM THIS DATE:
Potatoes may increase risk of type 2 diabetes—depending on their preparation
2025-08-06
Embargoed for release: Wednesday, August 6, 6:30 PM ET
Key points:
In a study that tracked the diets of more than 205,000 adults over decades, three servings weekly of French fries was associated with a 20% excess risk of developing type 2 diabetes (T2D). There was no significant association between consumption of baked, boiled, or mashed potatoes and risk of diabetes. However, swapping any form of potatoes for whole grains was estimated to lower T2D risk.
The study offers the most comprehensive findings on potatoes and T2D risk to date. Prior studies have ...
Three weekly servings of French fries linked to higher diabetes risk
2025-08-06
Eating three servings of French fries a week is associated with a 20% increased risk of developing type 2 diabetes, but eating similar amounts of potatoes cooked in other ways - boiled, baked or mashed - does not substantially increase the risk, finds a study published by The BMJ today.
What’s more, replacing any form of potatoes with whole grains was associated with a lower type 2 diabetes risk, but swapping them for white rice was linked to an increased risk, the results show.
Potatoes contain several nutrients including fibre, vitamin C, and magnesium, but they also have a high starch content and therefore a high glycemic index, so have been linked to a ...
Global hunt for ‘positive tipping points’
2025-08-06
Experts are calling for a global effort to identify “positive tipping points” to accelerate the green transition – and have devised a method to find them.
A tipping point is a threshold where a small change can trigger a significant, often irreversible, transformation.
With climate change becoming increasingly damaging and dangerous, positive tipping points in human societies and economies can spark rapid reductions in emissions and other environmental harm.
Positive tipping points are already happening – with solar power and electric vehicles rapidly displacing fossil-fuelled alternatives – but many more are needed to ensure a thriving, sustainable ...
Getting the most out of therapy – Therapists report what you should know before starting
2025-08-06
Of 1.7 million referrals for therapy in 2022-23, more than 1 million people dropped out before starting or during their therapy. This represents a dropout rate of over 60%. Many patients report worries about therapy, uncertainty about the process, and ill-preparedness for the journey ahead as to why they didn’t engage.
Using the scientific Delphi Consensus approach with 41 therapists, researchers from Warwick Medical School, University of Warwick, have published a report in BMJ Open listing recommendations for things people should know, and steps they can try, while awaiting therapy ...
Clean energy is here. Getting it to EVs isn’t
2025-08-06
Even if enough renewable energy is available, simply switching from gas-powered cars to electric vehicles (EVs) won’t be enough to fight climate change — unless the U.S. also upgrades its transmission grid, a new Northwestern University study finds.
If every gas-powered vehicle in the U.S. were replaced by an EV, transmission constraints would prevent the cleanest available electricity from reaching many charging locations. This “grid congestion” would force greater reliance on nearby fossil fuel power plants, undercutting the emissions benefits of electrification.
After identifying the issue, the study also recommends a modest set of targeted ...
Study: Affordable trial programs prevented youth substance misuse
2025-08-06
While overdose deaths in the U.S. sharply declined in 2024, they remain high. Almost 90,000 Americans died from drug overdoses between October 2023 and September 2024. Overdose death rates are particularly high in American Indian and Alaska Native (AI/AN) people.
The earlier someone starts substance use, the more likely they are to have substance use problems later in life. So, it is important to work with young people to prevent substance use early in life. Researchers at Emory’s Rollins School of Public Health recently partnered ...
Better access to technology can help African Americans bridge the healthcare gap
2025-08-06
(Santa Barbara, Calif.) — Researchers at UC Santa Barbara have found that better access to technology can go a long way toward achieving equity in healthcare for African Americans in the United States. According to Sharon Tettegah and Ebenezer Larnyo, both researchers at UCSB’s Center for Black Studies Research (CBSR), leveraging health technology is a promising strategy for addressing longstanding healthcare disparities “by enhancing access to care, improving its quality and reducing inequities.”
“Our study shows that improving access to technology like broadband internet and smartphones, and healthcare technology use like ...
Higher risk of ischemic stroke at young age after pregnancy complications
2025-08-06
The study examined over a thousand women aged 18 to 49. Of these, 358 had suffered an ischemic stroke. Researchers compared the pregnancy histories of these women with those of 714 women who had not experienced a stroke. The results showed that more than half of the women who had an ischemic stroke had experienced at least one complication during pregnancy. In the group without ischemic stroke, this was one-third.
Women who had experienced a stillbirth were found to be nearly five times more likely to ...
Complicated pregnancies linked to higher risk of early stroke
2025-08-06
EMBARGOED FOR RELEASE UNTIL 4:00 P.M. ET, WEDNESDAY, AUGUST 6, 2025
MINNEAPOLIS – Experiencing complications during pregnancy is linked with a higher risk of stroke before age 50, according to a study published August 6, 2025, in Neurology®, the medical journal of the American Academy of Neurology. Researchers found female participants who had certain pregnancy complications were more likely to have an early stroke. These included preeclampsia, preterm birth, gestational diabetes, miscarriage and stillbirth. The study does not prove that pregnancy complications cause stroke. It only shows an association.
“While the ...
American Society of Anesthesiologists hosts ANESTHESIOLOGY® 2025
2025-08-06
CHICAGO – Older patients whose primary language is not English may be at greater risk for delirium after surgery in American hospitals. People who struggle with food insecurity or loneliness may be more likely to experience chronic pain after surgery. Fentanyl overdose deaths in older adults are rising at similar rates to younger people. When anesthesiologists lead preoperative care for hip fractures, patients get to the operating room faster and recover better. These findings and other timely research ...